After the World Health Organization (WHO) has increased in June 2009 the pandemic alert for the new A type H1N1 virus (swine flu) to the highest level 6, there was no shortage of articles in the Internet and the press with empirical observations of the flu pandemic development. However, it was not possible back in May 2009 to locate any publicly available forecast of the likely future developments. To fill this gap Alex Borissov, MA MBA ACMA with first degree major in statistical analysis, has developed a forecast model predicting future number of H1N1 reported cases and deaths from June 1st, 2009 to the end of 2009 year.
The model is based on statistical observations of the dynamic of the laboratory confirmed H1N1 cases as reported by the World Health Organization. For the first month of running (June 2009) the forecast of the reported number of swine flu cases proved to be remarkably accurate.
It was observed that the 10 day average of daily increases in the number of cases as at the end of May 2009 was about 1.05% with the mortality rate of about 0.66%. Based on the total (for the World) number of cases reported by the WHO as at June 1st 2009 of 17,410, it was predicted that by July 1st the total number of cases will reach 75,659. In fact it was reported by the WHO at the level of 77,201 – i.e. with the error of 2%. Given the infrequent pattern of reporting by countries, with some countries like USD reporting on a weekly basis, the daily errors resulted in the average monthly error of 0.37%, as follows:
|
In a nutshell, the extrapolation of the present rate of increase of H1N1 reported cases to future periods suggests that during the period between January 15th, 2010 (worst case scenario) and February 19th, 2010 (best case scenario) every person on planet Earth will be reported as having had H1N1 symptoms between April 2008 and the then present time.
The above model has several imbedded limitations. It could be accurate in the short run while the assumptions remain in place, but will become inaccurate in the longer term.
The model is based on the assumption that the average rate of increase of the reported cases remains stable. This assumption can be made with reason for a short period of time and in the long term it is most likely to be incorrect. To mitigate this weakness we are planning to monitor the averages and provide further re-forecasts when the deviation proves to be significant.
The General guidance for all countries, published by WTO in the Weekly Epidemiological Record, No. 26, volume 84, June 26, 2009 advised that although the countries should continue monitoring disease and outbreaks to assess whether the levels of influenza activity are increasing or decreasing, in the countries with widespread community-level transmission the collection of clinical samples for virus testing should be limited to selected samples of cases only, rather then laboratory testing of all cases to save on resources. Therefore with the increase of the number of sick with influenza H1N1 over a certain level, the number of new reported laboratory confirmed cases will become less representative of the real number of actually sick.
In determining the actual number of sick with influenza H1N1 world wide, the number of the laboratory confirmed cases has limited utility. As advised by the WHO in the Weekly Epidemiological Record, No. 24, volume 84, June 12, 2009, point-of-care rapid diagnostic tests for influenza failed to correctly diagnose about 50% of cases in samples collected on the day of onset of illness. Further, as reported by Reuters on June 26th, on the day when the number of confirmed cases of H1N1 in the USA was reported at the level of 21,449, the U.S. health officials revealed that in reality as many as 1 million people in the United States may have been infected with the new H1N1 swine flu virus.
Therefore, taking into account the above notes, the actual number of sick with swine flu world wide may be running at present time into millions of people. In practical terms this means that the vaccination, when the vaccine becomes available in a couple of months from now as presently estimated, will have a limited impact on the disease, as by that time it will be much more widely spread than the present model suggests.
|
